CN102857755B - The method and apparatus determining transform block size - Google Patents

Abstract

The embodiment of the present invention provides a kind of method and apparatus determining transform block size, it is possible to increase picture compression efficiency.The method includes: according to image block information and the dividing mode of image block, it is thus achieved that the parameter of the transform block that image block is corresponding；Parameter according to transform block and the dividing mode of image block, it is thus achieved that transform block size.The embodiment of the present invention, when determining transform block size, according to parameter and the dividing mode of image block of transform block corresponding to image block, it is thus achieved that transform block size, therefore, it is possible to use the transform block of the dividing mode adapting to image block, improves picture compression efficiency.

Description

Method and device for determining transform block size

technical field

The invention relates to the field of communication technology, in particular to a method and device for determining the size of a transformation block.

Background technique

In order to minimize the bandwidth required to transmit video data, multiple video compression methods can be used to compress video data, wherein the video compression methods include: intra-frame compression and inter-frame compression. Nowadays, inter-frame compression methods based on motion estimation are mostly used. Specifically, the process of image compression and encoding by the image encoding end using the inter-frame compression method includes: the encoding end divides the image block to be encoded into several sub-image blocks of equal size, and then for each sub-image block, in the reference image Search for the image block that best matches the current sub-image block as the prediction block, then subtract the sub-image block from the corresponding pixel value of the prediction block to obtain a residual, and perform entropy coding on the value obtained by transforming and quantizing the residual , and finally send the bit stream and motion vector information obtained by entropy encoding to the decoding end, wherein the motion vector information represents the position difference between the current sub-image block and the prediction block. At the decoding end of the image, first obtain the entropy coded bit stream and perform entropy decoding to obtain the corresponding residual and the corresponding motion vector information; then obtain the corresponding matching image block (that is, the above-mentioned prediction block) in the reference image according to the motion vector information ), and then add the value of each pixel in the matching image block to the value of the corresponding pixel in the residual value to obtain the value of each pixel in the current sub-image block. Intra-frame prediction refers to using the information in the image to predict the image block to obtain the prediction block. The encoding end obtains the corresponding pixels of the prediction block according to the prediction mode, prediction direction, and pixel values around the image block, and compares the pixels of the image block with the prediction block The pixel is subtracted to obtain the residual, and the residual is written into the code stream after transformation, quantization and entropy coding; the decoding end parses the code stream, performs entropy decoding, inverse quantization, and inverse transformation on the code stream to obtain a residual block, and the decoding end according to The prediction mode, prediction direction, and pixel values around the image block are used to obtain a prediction block, and the pixels of the residual block and the prediction block are added to obtain a reconstructed image block.

In current video codec standards, there are concepts of coding unit, prediction unit and transform unit. Wherein, the coding unit is an image block that is operated when coding or decoding is performed at the coding end or the decoding end. A prediction unit is an image block with an independent prediction mode in a coding unit. A prediction block is an image block on which a coding unit performs a prediction operation, and one prediction unit may contain multiple prediction blocks. A transformation unit is an image block that undergoes transformation operations in a coding unit, and the image block at this time may also be called a transformation block. Considering that the correlation of the internal difference signal of the prediction block is strong, the energy concentration performance of large block transformation is higher than that of small block transformation. In a more general sense, an image block can contain one or more predictive blocks, which are predicted in units of predictive blocks at the codec end; at the same time, an image block contains one or more transform blocks, and are predicted in units of transform blocks at the codec end Make a transformation.

In existing video codec standards, such as Moving Picture Experts Group (MPEG), H.264/AVC (Advanced Video Coding, Enhanced Video Coding), an image block, or macroblock (macroblock) , super-macroblock (super-macroblock), etc., are divided into several sub-image blocks, and the size of these sub-image blocks can be 64×64, 64×32, 32×64, 32×32, 32×16, 16×32 , 16×16, 16×8, 8×16, 8×8, 8×4, 4×8, 4×4, etc., sub-image blocks perform the above-mentioned motion estimation and motion compensation with these sizes, and the encoding side of the image needs The code word identifying the division method of the image block is sent to the decoder of the image, so that the decoder of the image can know the division method of the image encoder, and determine the corresponding prediction block according to the division method and motion vector information. In the existing video codec standard, these sub-image blocks are rectangular blocks of N×M (both N and M are integers greater than 0), and N and M have a multiple relationship.

In the existing video coding and decoding technology, the transformation matrix can be used to remove the residual correlation of the image block, that is, to remove the redundant information of the image block, so as to improve the coding efficiency. The transformation of the data block in the image block usually adopts two Dimensional transformation, that is, the residual information of the data block is multiplied by an N×M transformation matrix and its transpose matrix at the encoding end, and the transformation coefficients are obtained after the multiplication. The above steps can be described using the following formula:

f=T'×C×T

Wherein, C represents the residual information of the data block, T and T' represent the transformation matrix and the transpose matrix of the transformation matrix, and f represents the transformation coefficient matrix obtained after the transformation of the residual information of the data block. Wherein, the transformation matrix may be a discrete cosine transform (Discrete Cosine Transform, DCT) matrix, an integer transform (Integer Transform) matrix, a KL transform (Karhunen Lòeve Transform, KLT) matrix, and the like. Among them, KLT can better consider the texture information of image blocks or image block residuals, so using KLT can achieve better results.

Performing the above processing on the residual information of the image block is equivalent to converting the residual information of the image block from the spatial domain to the frequency domain, and the transformation coefficient matrix f obtained after processing is concentrated in the low-frequency region; After the above transformation is performed, quantization, entropy coding and other processing are performed on the transformation coefficient matrix obtained after transformation, and the bit stream obtained by entropy coding is sent to the decoding end. In order for the decoding end to know the type and size of the transformation matrix used by the encoding end, usually, the encoding end will send the indication information indicating the transformation matrix used by the current image block to the decoding end.

The subsequent decoding end determines the transformation matrix used by the encoding end according to the above indication information, and according to the characteristics of the transformation matrix (orthogonality of the transformation matrix, etc.), decodes the bit stream sent by the encoding end to obtain the transformation coefficient matrix, and combines the transformation coefficient matrix with the transformation matrix By multiplying the matrix and its transposed matrix, the residual information of the data block that is approximately consistent with the encoding end can be recovered. The above steps can be described using the following formula:

C=T×f×T′

Wherein, C represents the residual information of the data block, T and T′ represent the transformation matrix and the transpose matrix of the transformation matrix, and f represents the transformation coefficient matrix obtained by the decoding end.

Since the residuals of image blocks may have different distribution rules, using a transformation matrix of a certain size often fails to achieve a good transformation effect. Therefore, in the prior art, attempts are made to use transformation matrices of different sizes for the residuals of image blocks (also known as a transform block). Therefore, for a 2N×2N image block, a transformation matrix with a size of 2N×2N may be used, or a transformation matrix with a size of N×N, or a transformation matrix with a size of 0.5N×0.5N may be used.

However, at present, only the transformation matrix of square size is used. For the frequently appearing strip texture, the transformation matrix of square size cannot effectively remove the redundant information of the image block, which reduces the image compression efficiency.

Contents of the invention

Embodiments of the present invention provide a method and device for determining a transformation block size, which can improve image compression efficiency.

On the one hand, a method for determining the size of a transformation block is provided, including: obtaining the parameters of the transformation block corresponding to the image block according to the image block information and the division method of the image block; according to the parameters of the transformation block and the division method of the image block, obtaining Transform block size.

In another aspect, an image coding method is provided, including: obtaining at least one candidate transform block size according to parameters of a transform block corresponding to the image block; determining the number of at least one candidate transform block size; selecting one of the candidate transform block sizes, As the size of the transform block corresponding to the image block.

On the other hand, an image decoding method is provided, including: obtaining at least one candidate transform block size according to the parameters of the transform block corresponding to the image block; determining the serial number of at least one candidate transform block size; obtaining the transform block size of the transform block Number; obtain the transformation block size according to the number of the transformation block size.

In another aspect, a device for determining the size of a transformation block is provided, including: a parameter acquisition unit, configured to obtain parameters of a transformation block corresponding to the image block according to image block information and a division method of the image block; a size acquisition unit, configured to The size of the transform block is obtained according to the parameters of the transform block and the division manner of the image block.

In another aspect, an image coding device is provided, including: an obtaining unit, configured to obtain at least one candidate transform block size according to a parameter of a transform block corresponding to the image block; a determining unit, configured to determine the size of at least one candidate transform block number; the coding unit is used to select one of the candidate transform block sizes as the size of the transform block corresponding to the image block, and encode the number of the size of the transform block.

In another aspect, an image decoding device is provided, including: an obtaining unit, configured to obtain at least one candidate transform block size according to a parameter of a transform block corresponding to the image block; a determining unit, configured to determine the size of at least one candidate transform block number; a decoding unit configured to obtain the number of the transform block size of the transform block, and obtain the transform block size according to the number of the transform block size.

In the embodiment of the present invention, when determining the size of the transformation block, the size of the transformation block is obtained according to the parameters of the transformation block corresponding to the image block and the division method of the image block, so the transformation block suitable for the division method of the image block can be used, and the image compression efficiency is improved. .

Description of drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some of the present invention. Embodiments, for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without any creative effort.

1a-1d are schematic diagrams of examples of symmetrical division methods of image blocks.

2a-2d are schematic diagrams of examples of asymmetric division of image blocks.

3a-3c are schematic diagrams of transform blocks corresponding to different layers.

Fig. 4a-Fig. 4d are schematic diagrams of determining the transform block size according to the prediction block type.

Fig. 5 is a flowchart of a method for determining a transform block size according to an embodiment of the present invention.

Fig. 6a and Fig. 6b are schematic diagrams of transform blocks according to an embodiment of the present invention.

7a-7d are schematic diagrams of transform blocks according to embodiments of the present invention.

Fig. 8 is a flowchart of an image coding method according to an embodiment of the present invention.

Fig. 9 is a flowchart of an image decoding method according to an embodiment of the present invention.

Fig. 10 is a block diagram of an apparatus for determining a transform block size according to an embodiment of the present invention.

Fig. 11 is a block diagram of an image encoding device of one embodiment of the present invention.

Fig. 12 is a block diagram of an image decoding device according to an embodiment of the present invention

detailed description

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

1a-1d are schematic diagrams of examples of symmetrical division methods of image blocks.

The commonly used ways to divide an image block into sub-image blocks are: 2N×2N division method, the image block only contains one sub-image block, that is, the image block is not divided into smaller sub-image blocks, as shown in Figure 1a; 2N×2N The N division method divides the image block into two sub-image blocks of equal size, as shown in Figure 1b; the N×2N division method divides the image block into two sub-image blocks of equal size on the left and right, as shown in Figure 1c Shown; N×N division method, the image block is divided into four sub-image blocks of equal size, as shown in Figure 1d. The above N is any positive integer, representing the number of pixels.

2a-2d are schematic diagrams of examples of asymmetric division of image blocks.

The image blocks may also be divided in an asymmetric manner, as shown in Fig. 2a to Fig. 2d. The division method shown in Figure 2a and 2b divides an image block into two unequal rectangular sub-image blocks up and down, in the two sub-image blocks divided by 2N*nU (n=0.5N in the figure) shown in Figure 2a, The lengths of the two upper sub-image blocks are 2N and 0.5N, and the lengths of the two lower sub-image blocks are 2N and 1.5N. Generally speaking, U in 2N×nU indicates that the image dividing line moves relative to the vertical bisector of the image block, and 2N×nU indicates that the image dividing line moves n relative to the vertical bisector of the image block, where n=x× N, where x is greater than or equal to 0 and less than or equal to 1.

Among the two sub-image blocks divided by 2N×nD (n=0.5N in the figure) shown in Figure 2b, the two sub-image blocks on the upper side become 2N and 1.5N in length, and the two sub-image blocks on the lower side become 2N and 0.5N in length N. Generally speaking, D in 2N×nD means that the image dividing line moves down relative to the vertical bisector of the image block, and 2N×nD means that the image dividing line moves down by n relative to the vertical bisector of the image block, n=x×N, where x is greater than or equal to 0 and less than or equal to 1.

The division methods shown in Figures 2c and 2d divide an image block into two unequal rectangular sub-image blocks on the left and right, in the two sub-image blocks divided by nL*2N (n=0.5N in the figure) shown in Figure 2c, The lengths of the two sub-image blocks on the left are 0.5N and 2N, and the lengths of the two sub-image blocks on the right are 1.5N and 2N. Generally speaking, L in nL×2N means that the image dividing line moves left relative to the vertical bisector of the image block, and nL×2N means that the image dividing line moves left by n relative to the vertical bisector of the image block, where n=x ×N, where x is greater than or equal to 0 and less than or equal to 1.

Among the two sub-image blocks divided by nR×2N (n=0.5N in the figure;) shown in Fig. 2d, the lengths of the two sub-image blocks on the left are 1.5N and 2N, and the lengths of the two sub-image blocks on the right are 0.5N and 2N. Generally speaking, R in nR×2N means that the image dividing line moves right relative to the vertical bisector of the image block, and nR×2N means that the image dividing line moves right by n relative to the vertical bisector of the image block, where n=x ×N, where x is greater than or equal to 0 and less than or equal to 1.

The foregoing manner of dividing an image block may also be referred to by a prediction block type. 2N×2N, 2N×N, N×2N, 2N×nU, 2N×nD, nL×2N, and nR×2N all represent prediction block types corresponding to image block division methods.

In order to effectively represent how image blocks use transformation matrices of different sizes, a tree identification method can be used to identify transformation blocks layer by layer. 3a-3c are schematic diagrams of transform blocks corresponding to different layers.

As shown in Figure 3a-Figure 3c, when identifying the transformation block size used by the image block, the zeroth layer in the code stream has an indicator bit for identifying whether the image block uses a transformation matrix with a size of 2N×2N, if the image block uses a size of 2N×2N transformation matrix (as shown in FIG. 3 a ), the indicator bit is 0. If the image block does not use 2N×2N transformation, the indicator bit is 1, indicating that the transformation matrix with a size of 2N×2N needs to be further divided into four transformation matrices with a size of N×N, and in the first layer structure of the code stream In , 4 bits are used to respectively identify whether each transformation matrix with a size of N×N is further divided. If the image block uses the transformation structure as shown in FIG. 3b, all 4 bits are 0, indicating that each transformation matrix with a size of N×N will not be further divided.

When the transformation structure shown in Figure 3c is selected, 2 bits out of 4 bits are 0, and 2 bits are 1. The 2 bits are 0, which means that the transformation matrix of size N×N in the lower left and upper right is not Then divide. Two bits being 1 indicate that the upper-left and lower-right transformation matrices with a size of N×N need to be further divided to obtain a transformation matrix with a size of 0.5N×0.5N. Then in the second layer structure of the code stream, use 4 bits to indicate whether to further divide the transformation matrix with a size of 0.5N×0.5N on the upper left, and use 4 bits to indicate whether to further divide the transformation matrix with a size of 0.5N×0.5 on the lower right The transformation matrix of N is further divided. If the image block uses the transformation structure as shown in FIG. 3c, the above 4+4 bits are all 0, indicating that no further division will be made. Through the above layer-by-layer identification in the code stream, the transformation block size used by the image block and the sub-image block can be effectively and flexibly indicated.

In addition to the above method of identifying the layer number and size corresponding to the transformation block layer by layer, there is also a method of determining the size of the transformation block according to the prediction block type used by the image block. Different prediction block types correspond to different image block division methods, and the prediction block can be obtained according to the image block division methods. For example, the 2N×N division means that the size of the corresponding prediction block is 2N×N (as shown in Figure 1b).

In video coding and decoding technology, usually there will be a jumpy transformation in the residual data corresponding to the boundary of two prediction blocks, so if the transformation matrix crosses the boundary of two prediction blocks, the effect of transformation will be weakened and cannot be effectively Removing the residual correlation of the image block cannot effectively remove the redundant information of the image block and reduce the coding efficiency. Fig. 4a-Fig. 4d are schematic diagrams of determining the transform block size according to the prediction block type.

As shown in Figure 4a-Figure 4d, when the image blocks respectively use 2N×2N, 2N×N, N×2N, N×N prediction block types, the corresponding transformation blocks of the image are divided into 2N×2N, N×N , N×N, N×N conversion. The purpose of this method is to ensure that the transformation block does not cross the boundary of the prediction block, so as to ensure the coding efficiency. Therefore, when the image block uses 2NxN, Nx2N, NxN prediction block types, the image uses four NxN transform blocks. This approach is often referred to as an implicit split.

In methods using layer-wise identification and implicit partitioning, the size of the transformation matrix is not linked to the size of the prediction block. As shown in Figure 4b, when the 2N×2N image block uses the 2N×N division method, since the division method reflects the texture information of the area where the image block is located, the texture of the area where the image block is located is more inclined to have a horizontal texture. features, but at this time the image block still uses N×N transform blocks.

Since the size of the transformation matrix is not related to the size of the prediction block, the transformation matrix does not use the information of the prediction block to effectively remove the redundant information of the image block, thereby affecting the coding efficiency. In the embodiment of the present invention, when determining the size of the transform block, the transform block adapted to the division mode is used, so that the image compression efficiency can be improved.

Fig. 5 is a flowchart of a method for determining a transform block size according to an embodiment of the present invention.

501. Obtain parameters of a transformation block corresponding to the image block according to the image block information and the division manner of the image block.

Optionally, in an embodiment, the above parameters of the transformation block include the number of layers corresponding to the transformation block (hereinafter referred to as trafoDepth) or the representation value of the size of the transformation block (hereinafter referred to as log2TrafoSize).

For example, the relationship between the transform block size and the representation value of the transform block size can be represented by the following equation:

Transform block size = 2 ^{log2TrafoSize} , or it can be expressed as 1<<log2TrafoSize by shifting method. Among them, "<<" represents a shift operation to the left.

Optionally, in another embodiment, the image block information may include the image block size, the representation value of the image block size, the number of layers of the image block, or the serial number of the image block size. The forms of these image block information correspond to each other, and the representation value of the image block size (hereinafter referred to as log2CUSize) can be obtained directly or indirectly. For example, the relationship between the image block size and the representation value of the image block size can be expressed by the following formula:

Image block size = 2 ^log2CUSize , or it can be expressed as 1<<log2CUSize by shifting method. Among them, "<<" represents a shift operation to the left.

In the following, for the sake of brevity, the image block size representation value log2CUSize is used as an example of the image block information, but the embodiment of the present invention is not limited thereto, and the image block information may include other forms.

Optionally, in an embodiment, the representation value of the transform block size is equal to the representation value of the image block size minus the number of layers corresponding to the transform block, log2TrafoSize=log2CUSize-trafoDepth.

Alternatively, the representation value of the transform block size is equal to the representation value of the image block size minus m1, where m1 is a positive integer, log2TrafoSize=log2CUSize-m1.

Alternatively, the representation value of the transform block size is equal to the representation value of the image block size minus the number of layers corresponding to the transform block minus m2, m2 is a positive integer, log2TrafoSize=log2CUSize-trafoDepth-m2.

In addition, the prediction block can be obtained according to the division manner of the image block.

502. Obtain the size of the transform block according to the parameters of the transform block and the division manner of the image block.

Optionally, in an embodiment, the first size of the transform block may be obtained according to parameters of the transform block. For example, when the parameter is the layer number trafoDepth corresponding to the transform block, the first size=2 ^{log2CUSize-trafoDepth} or the first size=1<<(log2TrafoSize-trafoDepth). Or, when the parameter is the representative value log2TrafoSize of the transform block size, the first size=2 ^{log2TrafoSize} or the first size=1<<log2TrafoSize.

When determining the size of the transform block, the size of the transform block is determined according to the first size. For example, the horizontal size of the transform block is equal to h times the first size, and the vertical size of the transform block is equal to v times the first size, where h and v are greater than 0. For example, when the horizontal size of the predicted block is larger than the vertical size, h is greater than v, or when the horizontal size of the predicted block is smaller than the vertical size, h is smaller than v, or when the horizontal size of the predicted block is equal to the vertical size In the direction dimension, h is equal to v.

Embodiments of the present invention When determining the size of the transformation block, the embodiment of the present invention obtains the size of the transformation block according to the parameters of the transformation block corresponding to the image block and the division method of the image block. Therefore, a transformation block suitable for the division method of the image block can be used to improve image compression efficiency.

The method in Fig. 5 can be used in the process of image encoding and decoding. For example, during the encoding process, the transform block size obtained by using the method described in FIG. 5 may be encoded, and the encoded result may be written into the code stream. On the other hand, in the decoding process, the above encoding result can be analyzed from the code stream, and the corresponding transform block size can be obtained according to the method described in FIG. 5 .

The embodiments of the present invention will be described in more detail below in conjunction with specific examples.

Fig. 6a and Fig. 6b are schematic diagrams of transform blocks according to an embodiment of the present invention. This embodiment is described below in conjunction with FIGS. 1a-1d.

In this embodiment, the layer number trafoDepth corresponding to the transform block is used as a parameter of the transform block. Firstly, the size of the image block is determined to be 2M×2M, and 2M is represented by the parameter log2CUSize, and the relationship between the two is 2M=1<<log2CUSize.

Then, the number of layers trafoDepth where the transformation block corresponding to the image block is located is determined according to the division manner of the image block. As shown in Figure 1a, the image block is divided by 2N×2N, and the division method is 2N×2N. At this time, the transformation block corresponding to the image block is located at the zeroth layer, and the value of trafoDepth is 0. As shown in FIG. 1b, the image block is divided by 2N×N, and the division method is 2N×N. At this time, the transformation block corresponding to the image block is located in the first layer, and the value of trafoDepth is 1. As shown in FIG. 1c, the image block is divided by N×2N, and the division method is N×2N. At this time, the transformation block corresponding to the image block is located in the first layer, and the value of trafoDepth is 1. As shown in FIG. 1d, the image block is divided by N×N, and the division method is N×N. At this time, the transformation block corresponding to the image block is located in the first layer, and the value of trafoDepth is 1. At this time, the first size is 1<<(log2CUSize-trafoDepth).

The length and width of the transform block are determined according to the layer number of the transform block corresponding to the determined image block and the division method used for the image block. As shown in Figure 1a, the image block is divided into 2N×2N. At this time, the number of layers corresponding to the transformation block is 0, and the size of the transformation block corresponding to the image block is consistent with the size of the image block, which is still 2M×2M (h=2, v =2), that is, the transform block size is (1<<log2CUSize)×(1<<log2CUSize ).

As shown in FIG. 1 b , the image block is divided in a 2N×N manner, and the number of layers corresponding to the transform block is 1 at this time. At this time, the image texture corresponding to the region where the image block is located is more inclined to the horizontal image texture, and using 2N×0.5N transform blocks can improve coding efficiency more than N×N transform blocks.

In this embodiment, when the image block is divided into 2N×N, the transformation block uses 2M×0.5M transformation (h=2, v=0.5) as shown in Figure 6a, that is, the horizontal dimension of the transformation block and the horizontal dimension of the image block Consistently, the vertical size of the transformation block is a quarter of the vertical size of the image block. If expressed by the parameters log2CUSize and trafoDepth, the transform block size corresponding to the 2N×N division method is (1<<log2CUSize)×(1<<(log2CUSize-trafoDepth-1)), or (1<<log2CUSize)×( (1<<log2CUSize)>>(trafoDepth+1)).

Similarly, when the image block uses the N×2N division method as shown in Figure 1c, the number of layers corresponding to the transformation block is 1, and the image texture corresponding to the area where the image block is located is more inclined to the vertical image texture, use 0.5M× A 2M transform block can improve coding efficiency more than an N×N transform block. Therefore, when the image block uses the N×2N division method in the present invention, the transformation block uses 0.5M×2M transformation (h=0.5, v=2) as shown in Figure 6b, that is, the horizontal dimension of the transformation block is the horizontal direction of the image block A quarter of the size, the vertical size of the transformation block is consistent with the vertical size of the image block. If expressed by the parameters log2CUSize and trafoDepth, the transform block size corresponding to the N×2N division method is (1<<(log2CUSize-trafoDepth-1))×(1<<log2CUSize), or ((1<<log2CUSize)> >(trafoDepth+1))×(1<<log2CUSize).

As shown in Figure 1d, the image block uses the N×N division method, and the number of layers corresponding to the transformation block is 1. Due to the N×N division method, the 2N×2N image block is divided into four equal-sized square predictions along the horizontal and vertical directions. block, it is impossible to judge the texture information of the area where the image block is located according to the size of the predicted block, so the size of the transformed block corresponding to the N×N predicted block is M×M (h=1, v=1), that is, the horizontal direction and vertical direction of the transformed block The orientation dimensions are each half the size of the image block. Indicated by the parameters log2CUSize and trafoDepth, the transform block size corresponding to the N×N division method is (1<<(log2CUSize-trafoDepth))×(1<<(log2CUSize-trafoDepth)), or ((1<<log2CUSize) >>trafoDepth)×((1<<log2CUSize)>>trafoDepth).

Specifically, when the size of the image block is 32×32, as described above, the value of trafoDepth corresponding to the layer number trafoDepth of the transformation block divided into 2N×2N, 2N×N, N×2N, and N×N is respectively 0. .

In the embodiment of the present invention, when determining the size of the transformation block, the size of the transformation block is obtained according to the parameters of the transformation block corresponding to the image block and the division method of the image block, so the transformation block suitable for the division method of the image block can be used, and the image compression efficiency is improved. .

Another embodiment of the present invention is described below. In this embodiment, the parameter of the transformation block is log2TrafoSize, which represents the size of the transformation block.

Firstly, the size of the image block is determined to be 2M×2M, and 2M is represented by the parameter log2CUSize, and the relationship between the two is 2M=1<<log2CUSize. The initial value of log2TrafoSize is log2CUSize, and log2TrafoSize corresponding to different layers can be obtained according to log2TrafoSize=log2CUSize-trafoDepth.

Then determine the value of the transformation block log2TrafoSize corresponding to the image block according to the division method of the image block. As shown in Figure 1a, the image block is divided by 2N×2N, and the division method is 2N×2N. At this time, the value of log2TrafoSize of the transformation block corresponding to the image block remains unchanged, which is equal to the initial value of log2CUSize. As shown in FIG. 1b , the image block is divided by 2N×N, and the division method is 2N×N. At this time, the value of log2TrafoSize of the transformation block corresponding to the image block is reduced by one, that is, log2TrafoSize=log2CUSize-1. As shown in Figure 1c, the image block is divided by N×2N, and the division method is N×2N. At this time, the value of log2TrafoSize of the transformation block corresponding to the image block is reduced by 1, that is, log2TrafoSize=log2CUSize-1. As shown in FIG. 1d, the image block is divided by N×N, and the division method is N×N. At this time, the value of log2TrafoSize of the transformation block corresponding to the image block is reduced by one, that is, log2TrafoSize=log2CUSize-1.

Here, the above-mentioned log2TrafoSize=log2CUSize-1 may be based on the formula log2TrafoSize=log2CUSize-m1 (m1=1), or based on the formula log2TrafoSize=log2CUSize-trafoDepth, wherein, similar to that described in Embodiment 1, the division method is 2N×N, N× The number of layers corresponding to 2N, N×N transform blocks is trafoDepth=1.

According to the aforementioned principle, the length and width of the transform block are determined by the determined parameter log2TrafoSize of the transform block and the division method used for the image block. The length of the transform block refers to the size of the transform block in the horizontal direction, and the width of the transform block refers to the size of the transform block in the vertical direction.

At this time, the first size is 1<<log2TrafoSize. As shown in Figure 1a, the image block is divided into 2N×2N, and the size of the transformation block corresponding to the image block is consistent with the size of the image block, which is still 2M×2M (h=2, v=2), that is, the size of the transformation block at this time It is (1<<log2TrafoSize)×(1<<log2TrafoSize). As shown in Figure 1b, the image block uses 2N×N division method, and at this time, the transformation block uses 2M×0.5M transformation (h=2, v=0.5) as shown in Figure 5a, that is, the horizontal dimension of the transformation block is equal to the image block level The size of the direction is consistent, and the vertical size of the transformation block is a quarter of the vertical size of the image block. If it is expressed by the parameter log2TrafoSize, the size of the transformation block corresponding to the 2N×N division method is (1<<(log2TrafoSize+1)) ×(1<<(log2TrafoSize-1)). Similarly, when the image block uses the N×2N division method as shown in Figure 1c, the transformation block uses 0.5M×2M transformation (h=0.5, v=2) as shown in Figure 5b, that is, the horizontal dimension of the transformation block is A quarter of the horizontal size of the block, and the vertical size of the transformation block is consistent with the vertical size of the image block. With the parameter log2TrafoSize, the size of the transformation block corresponding to the N×2N division method is (1<<(log2TrafoSize-1)) ×(1<<(log2TrafoSize+1)).

As shown in Figure 1d, the image block is divided by N×N, and the size of the transformation block is M×M (h=1, v=1), that is, the horizontal and vertical dimensions of the transformation block are each half of the size of the image block 1. It is represented by the parameter log2TrafoSize. At this time, the transform block size corresponding to the N×N division method is (1<<log2TrafoSize)×(1<<log2TrafoSize).

Specifically, when the image block size is 32×32, as described above, the value of log2CUSize is 5, and the initial value of log2TrafoSize is 5. First, 2N×2N, 2N×N, N×2N, and N×N divisions are obtained The values of log2TrafoSize corresponding to the method are 5, 4, 4, and 4. According to the above division method, the transform block sizes are 32×32, 32×8, 8×32, and 16×16, respectively.

In the embodiment of the present invention, when determining the size of the transformation block, the size of the transformation block is obtained according to the parameters of the transformation block corresponding to the image block and the division method of the image block, so the transformation block suitable for the division method of the image block can be used, and the image compression efficiency is improved. .

Another embodiment of the present invention is described below. Similar to the previous embodiment, the size of the image block is determined to be 2M×2M, 2M is represented by the parameter log2CUSize, the relationship between the two is 2M=1<<log2CUSize, and the parameters used to identify the basic information of the transformed block size are log2TrafoSize, log2TrafoSize The initial value of log2CUSize. In this embodiment, before 501 in FIG. 5, the division flag value (hereinafter referred to as InterSplitFlag) is obtained according to the division method and the layer flag value (hereinafter referred to as max_transform_hierarchy_depth_inter), and the layer flag value is used to represent the transform block Whether to use layer-by-layer identification, and determine whether to determine the transformation block size according to the division method of the image block based on the division flag value.

In the video encoding and decoding method, an image block usually tries to use transformations of different sizes to transform the data of the image block, and the aforementioned layer-by-layer identification method is used to identify the layer number of the transformed block or the size of the transformed block. At the same time, in the video encoding and decoding method, an image block may only use a transformation block of one size to transform the image block data, that is, the above-mentioned layer-by-layer identification method is not used to identify the size of the transformation block or the number of layers where the transformation block is located. When the two methods are used simultaneously in the video codec standard or system, parameters are usually used to identify the two methods. Set this parameter to max_transform_hierarchy_depth_inter. When the encoding end marks the value of max_transform_hierarchy_depth_inter in the code stream as 1, it means that the image block at the encoding end uses transformation blocks of various sizes to transform the image block data, and uses the aforementioned layer-by-layer identification method to identify the transformation block The number of layers or the size of the transform block; when the encoder marks the max_transform_hierarchy_depth_inter value in the code stream as 0, it means that the image block at the encoder uses a transform block of one size to transform the image block data; when the decoder parses the code stream to get max_transform_hierarchy_depth_inter When the value is 1, it means that the image block at the decoding end uses the layer-by-layer identification method to obtain the number of layers where the transformation block is located or the size of the transformation block; when the value of maxt_ransform_hierarchy_depth_inter is 0 when the decoding end parses the code stream, it means that the image block at the decoding end uses a size The transform block transforms the image block data.

The method of determining the value of the transformation block log2TrafoSize corresponding to the image block according to the division method of the image block is as follows:

First get the value of the variable InterSplitFlag according to the table below:

InterSplitFlag How to divide image blocks 0 2N×2N ! (max_transform_hierarchy_depth_inter) 2N×N ! (max_transform_hierarchy_depth_inter) N×2N ! (max_transform_hierarchy_depth_inter) N×N ! (max_transform_hierarchy_depth_inter) 2N×nU ! (max_transform_hierarchy_depth_inter) 2N×nD ! (max_transform_hierarchy_depth_inter) nL×2N ! (max_transform_hierarchy_depth_inter) nR×2N

For example, the above 2N×nD, 2N×nU, nL×2N and nR×2N division methods respectively correspond to the divisions shown in Figures 2a, 2b, 2c and 2d, and n=0.5N.

When the value of InterSplitFlag is 0 and the value of max_transform_hierarchy_depth_inter is not 0, the value of log2TrafoSize can be determined according to the formula log2TraSfoize=log2CUSize-trafoDepth when using the aforementioned layer-by-layer identification method. For example, when the transformation block is on the zeroth layer, the value of log2TrafoSize remains unchanged, that is, log2TrafoSize=log2CUSize; when the transformation block is on the first layer, the value of log2TrafoSize relative to the zeroth layer log2TrafoSize is reduced by 1, that is, log2TrafoSize=log2CUSize- 1; when the transformation block is located in the second layer, the value of log2TrafoSize relative to the value of log2TrafoSize in the first layer is reduced by 1, that is, log2TrafoSize=log2CUSize-2; and so on to obtain the value of log2TrafoSize. When the value of InterSplitFlag is 0 and the value of max_transform_hierarchy_depth_inter is 0, use the aforementioned implicit split method (implicit split) to determine the transform block size, for example, when the value of max_transform_hierarchy_depth_inter is 0, the value of log2TrafoSize of the transform block corresponding to the 2N×2N partition type is a log2CUSize of size 2M by 2M.

When the value of InterSplitFlag is 1, as shown in Figure 1b, the image block is divided by 2N×N, and the division method is 2N×N. At this time, the value of the transformation block log2TrafoSize corresponding to the image block is reduced by 1, that is, log2TrafoSize=log2CUSize- 1. As shown in FIG. 1c, the image block is divided by N×2N, and the division method is N×2N. At this time, the value of log2TrafoSize of the transformation block corresponding to the image block is reduced by one, that is, log2TrafoSize=log2CUSize-1. As shown in FIG. 1d, the image block is divided by N×N, and the division method is N×N. At this time, the value of log2TrafoSize of the transformation block corresponding to the image block is reduced by one, that is, log2TrafoSize=log2CUSize-1. At this time, the first size is 1<<log2TrafoSize.

Then, the length and width of the transformation block are determined by the determined parameter log2TrafoSize of the transformation block and the division method used by the image block. As shown in Figure 1b, the image block uses a 2N×N division method. At this time, the transformation block uses a 2M×0.5M transformation (h=2, v=0.5) as shown in Figure 6a, that is, the horizontal dimension of the transformation block is the same as the image block level The size of the direction is consistent, and the vertical size of the transformation block is a quarter of the vertical size of the image block, expressed by the parameter log2TrafoSize. At this time, the size of the transformation block corresponding to the 2N×N division method is (1<<(log2TrafoSize+1))× (1<<(log2TrafoSize-1)). Similarly, when the image block uses the N×2N division method as shown in Figure 1c, the transformation block uses 0.5M×2M transformation (h=0.5, v=2) as shown in Figure 6b, that is, the horizontal dimension of the transformation block is the image A quarter of the horizontal size of the block, and the vertical size of the transformation block is consistent with the vertical size of the image block. With the parameter log2TrafoSize, the size of the transformation block corresponding to the N×2N division method is (1<<(log2TrafoSize-1)) ×(1<<(log2TrafoSize+1)). As shown in Figure 1d, the image block is divided by N×N, and the size of the transformation block is M×M (h=1, v=1), that is, the horizontal and vertical dimensions of the transformation block are each half of the size of the image block 1. It is represented by the parameter log2TrafoSize. At this time, the transform block size corresponding to the N×N division method is (1<<log2TrafoSize)×(1<<log2TrafoSize).

Specifically, when the image block size is 32×32, as described above, the value of log2CUSize is 5, the initial value of log2TrafoSize is 5, if the value of max_transform_hierarchy_depth_inter is 1, then 2N×N, N×2N, N×N The values of log2TraSfoize corresponding to the division method are 4, 4, and 4. According to the division method, the transform block sizes are respectively 32×8, 8×32, and 16×16.

When the value of InterSplitFlag is 1, as shown in Figure 2a-Figure 2d, the image blocks are divided into 2N×nU, 2N×nD, nL×2N and nR×2N respectively, and the division methods are 2N×nU, 2N×nD, nL×2N, and nR×2N. At this time, the value of log2TrafoSize of the transformation block corresponding to the image block is reduced by one, that is, log2TravSize=log2CUSize-1. Then, the length and width of the transformation block are determined by the determined parameter log2TrafoSize of the transformation block and the division method used by the image block. The transform block size corresponding to the 2N×nU and 2N×nD division types is (1<<(log2TrafoSize+1))×(1<<(log2TrafoSize-1)), that is, the transform block size is 2M×0.5M (h=2 , v=0.5). The transform block size corresponding to the nL×2N and nR×2N division types is (1<<(log2TrafoSize-1))×(1<<(log2TrafoSize+1)), that is, the transform block size is 0.5M×2M (h=0.5 , v=2).

Specifically, when the image block size is 32×32, as described above, the value of log2CUSize is 5, the initial value of log2TrafoSize is 5, if the value of max_transform_hierarchy_depth_inter is 1, then 2N×nU, 2N×nD, nL×2N The values of log2TrafoSize corresponding to the nR×2N division mode are 4, 4, 4, and 4. According to the division mode, the transform block sizes are obtained as 32×8, 32×8, 8×32, and 8×32, respectively.

In the embodiment of the present invention, when determining the size of the transformation block, the size of the transformation block is obtained according to the parameters of the transformation block corresponding to the image block and the division method of the image block, so the transformation block suitable for the division method of the image block can be used, and the image compression efficiency is improved. .

Another embodiment of the present invention is described below. Similar to the previous embodiment, the size of the image block is determined to be 2M×2M, 2M is represented by the parameter log2CUSize, the relationship between the two is 2M=1<<log2CUSize, and the parameters used to identify the basic information of the transformed block size are log2TrafoSize, log2TrafoSize The initial value of log2CUSize. In this embodiment, before 501 in FIG. 5, the division flag value (hereinafter referred to as InterSplitFlag) is obtained according to the division method and the layer flag value (hereinafter referred to as max_transform_hierarchy_depth_inter), and the layer flag value is used to represent the transform block Whether to use layer-by-layer identification, and determine whether to determine the transformation block size according to the division method of the image block based on the division flag value.

Let the number of division layers of the transformation block be trafoDepth, and the size of the transformation block is represented by log2TrafoSize, that is, the size of the transformation block is (1<<log2TrafoSize), and log2TrafoSize is the representation value of the transformation block size. The initial value of trafoDepth is 0.

When the encoding end marks the value of max_transform_hierarchy_depth_inter in the code stream as 1, it means that the image block at the encoding end uses transformation blocks of various sizes to transform the image block data, and uses the aforementioned layer-by-layer identification method to identify the number of layers or transformation blocks where the transformation block is located size; when the encoding end marks the max_transform_hierarchy_depth_inter value in the code stream as 0, it means that the image block at the encoding end uses a transformation block of one size to transform the image block data; when the decoding end parses the code stream and obtains a max_transform_hierarchy_depth_inter value of 1, it means The image block at the decoding end uses the layer-by-layer identification method to obtain the number of layers where the transform block is located or the size of the transform block; when the decoding end parses the code stream and obtains a max_transform_hierarchy_depth_inter value of 0, it means that the image block at the decoding end uses a transform block of one size for the image block The data is transformed.

Let the image block size be 2Mx2M. When the value of max_transform_hierarchy_depth_inter is 0, the image block uses a transform block of one size to transform the image block data. At this time, if the size of the transform block is still square (square, that is, the width and height of the transform block are equal), then when the image block is divided into 2NxN or Nx2N, the size of the transform block is MxM, which can ensure that the transform block does not span Over the prediction block boundary, the value of trafoDepth changes from 0 to 1 at this time. When the division method of the image block is 2NxnU or 2NxnD or nLx2N or nLx2N, the size of the transformation block is 0.5Mx0.5M to ensure that the transformation block will not cross the boundary of the prediction block. At this time, the value of trafoDepth changes from 0 to 2.

Specifically, in the codec operation, the size of the transform block can be determined according to the following steps.

(1) Obtain the division parameter interSplitFlag according to the division method of the image block and the number of layers of the transformation block. When the division method of the image block is 2NxN or Nx2N, if the number of layers of the transformation block is 0, it means that the size of the transformation block is consistent with the image block at this time. According to the aforementioned principle, it can be known that the transformation block still crosses the boundary of the prediction block, so the transformation block To divide into smaller values, set the value of interSplitFlag to 1 at this time, indicating that the transformation block needs to be divided; At this time, the size of the transformation block is the same as that of the image block. According to the aforementioned principle, the transformation block should be divided into smaller values. If the number of layers of the transformation block is 1, it means that the width and height of the transformation block are 1/2 of the image block. According to The aforementioned principle shows that the transform block still crosses the boundary of the prediction block, so when the layer number of the transform block is 0 or 1, the transform block needs to be divided. At this time, the value of interSplitFlag is 1, indicating that the transform block needs to be divided. Let the image block division mode be referred to by PartMode, and the division types of 2NxN, Nx2N, 2NxnU, 2NxnD, nLx2N, and nLx2N are represented by 2NxN, Nx2N, 2NxnU, 2NxnD, nLx2N, and nLx2N, respectively. The derivation process of interSplitFlag can be expressed by the following formula:

interSplitFlag=(max_transform_hierarchy_depth_inter==0&&PredMode==MODE_INTER&&((PartMode==PART_2NxN||PartMode==PART_Nx2N)&&trafoDepth==0)||((|Mode==PART_2NxnU|| PartMode==PART_nRx2N)&&trafoDepth<=1))

(2) Obtain the parameter interFurtherSplitFlag whether to further divide according to the division parameter interSplitFlag and the division method of the image block, and judge whether the transformation block is further divided, that is, whether the transformation block is divided into a size of 0.5Mx0.5M. When the value of interSplitFlag is 1 and the division mode of the image block is 2NxnU or 2NxnD or nLx2N or nLx2N, the value of interFurtherSplitFlag is 1. The derivation process of interFurtherSplitFlag can be expressed by the following formula:

interFurtherSplitFlag=interSplitFlag&&(PartMode==PART_2NxnU||PartMode==PART_2NxnD||PartMode==PART_nLx2N||PartMode==PART_nRx2N)

(3) Determine the maximum number of transform block layers corresponding to the transform block according to interSplitFlag or interSplitFlag and interFurtherSplitFlag parameters. When interSplitFlag is 1, the size of the transformation block changes from 2Mx2M to a smaller size, and the maximum number of transformation block layers is 1; when the value of interFurtherSplitFlag is 1, the transformation block needs to be divided into smaller sizes. The maximum number of transform block layers is 2. The maximum number of transform block layers can be derived from the following formula:

maxDepth=max_transform_hierarchy_depth_inter+interSplitFlag+interFurtherSplitFlag

(4) When the value of interSplitFlag is 1, it means that the transformation block needs to be divided, and the value of setting the transformation block division flag to 1 is used to indicate that the transformation block needs to be divided, where the transformation block division flag is represented by split_transform_flag

(5) When the value of split_transform_flag is 1, the number of layers of the transform block is increased by 1, that is, trafoDepth=trafoDepth+1; when the value of split_transform_flag is 1, the value of log2TrafoSize is reduced by 1, that is, log2TrafoSize=log2TrafoSize-1

(6) When the value of trafoDepth is greater than or equal to maxDepth, the encoder does not encode split_transform_flag;

(7) When the value of trafoDepth is greater than or equal to maxDepth, the decoder does not decode split_transform_flag.

Another embodiment of the present invention is described below. In this embodiment, first determine the size of the image block as 2M×2M, 2M is represented by the parameter log2CUSize, the relationship between the two is 2M=1<<log2CUSize, the parameter used to identify the basic information of the transformation block size is log2TrafoSize, the initial value of log2TrafoSize The value is log2CUSize. Determine the value of the transformation block log2TrafoSize corresponding to the image block according to the division method of the image block. As shown in Figures 6a and 6b, the image blocks are divided by 2N×nU and 2N×nD respectively, and the division methods are 2N×nU and 2N×nD. At this time, n=0.25, that is, according to the division method of the image block 2N×nU, the image block It is divided into two parts of 2M×0.25M and 2M×1.75M; according to the division method of image block 2N×nD, the image block is divided into two parts of 2M×1.75M and 2M×0.25M. In order to ensure that the transform block does not cross the prediction block boundary, the transform block size needs to be made smaller to fit the prediction block size. The value of log2TrafoSize is reduced by 2, that is, log2TrafoSize=log2CUSize-2; as shown in Figure 6c and 6d, the image block is divided by nL×2N and nR×2N respectively, and the division method is nL×2N and nR×2N, and n=0.25 at this time , that is, according to the division method of image block nL×2N, the image block is divided into two parts: 0.25M×2M and 2M×1.75M; according to the division method of image block nR×2N, the image block is divided into two parts: 1.75M×2M and 0.25M×2M part. In order to ensure that the transform block does not cross the prediction block boundary, the transform block size needs to be made smaller to fit the prediction block size. The value of log2TrafoSize is subtracted by 2, that is, log2TrafoSize=log2CUSize-2.

Here, the above-mentioned log2TrafoSize=log2CUSize-2 may be based on the formula log2TrafoSize=log2CUSize-m1 (m1=2), or based on the formula log2TrafoSize=log2CUSize-trafoDepth-m2, wherein, similar to that described in Embodiment 1, the division method is 2N×N, The number of layers corresponding to N×2N and N×N transform blocks is trafoDepth=1, and m2=1 in addition. Although m1=2 or m2=1 is taken as an example here, the embodiment of the present invention is not limited to these specific values, for example, m1 or m2 may take a larger value.

At this time, the first size is 1<<log2TrafoSize.

Then, the length and width of the transformation block are determined by the determined parameter log2TrafoSize of the transformation block and the division method used by the image block. As shown in Figures 7a and 7b, the transform block size corresponding to the 2N×nU and 2N×nD division methods is (1<<(log2TrafoSize+2))×(1<<(log2TrafoSize-1)), that is, the transform block size is 2M×0.25M (h=2, v=0.25). As shown in Figures 7c and 7d, the transform block size corresponding to the nL×2N and nR×2N division methods is (1<<(log2TrafoSize-1))×(1<<(log2TrafoSize+2)), that is, the transform block size is 0.25M×2M (h=0.25, v=2).

Specifically, when the image block size is 32×32, as described above, the value of log2CUSize is 5, and the initial value of log2TrafoSize is 5. First, 2N×nU, 2N×nD, nL×2N and nR×2N divisions are obtained The values of log2TrafoSize corresponding to the method are 3, 3, 3, and 3. According to the division method, the transformation block sizes are obtained as 32×4, 32×4, 4×32, and 4×32, respectively.

In the embodiment of the present invention, when determining the size of the transformation block, the size of the transformation block is obtained according to the parameters of the transformation block corresponding to the image block and the division method of the image block, so the transformation block suitable for the division method of the image block can be used, and the image compression efficiency is improved. .

Another embodiment of the present invention is described below. In this embodiment, first determine the size of the image block as 2M×2M, 2M is represented by the parameter log2CUSize, the relationship between the two is 2M=1<<log2CUSize, the parameter used to identify the basic information of the transformation block size is log2TrafoSize, the initial value of log2TrafoSize The value is log2CUSize. Determine the value of the transformation block log2TrafoSize corresponding to the image block according to the division method of the image block. As shown in Figures 7a and 7b, the image blocks are divided by 2N×nU and 2N×nD respectively, and the division methods are 2N×nU and 2N×nD. At this time, n=0.25, that is, according to the division method of the image block 2N×nU, the image block It is divided into two parts: 2M×0.25M and 2M×1.75M; according to the image block 2N×nD division method, the image block is divided into two parts: 2M×1.75M and 2M×0.25M. At this time, the value of log2TrafoSize is reduced by 1, that is, log2TrafoSize = log2CUSize-1.

As shown in Figures 7c and 7d, the image blocks are divided by nL×2N and nR×2N respectively, and the division methods are nL×2N and nR×2N. At this time, n=0.25, that is, according to the division method of the image block nL×2N, the image block It is divided into two parts of 0.25M×2M and 2M×1.75M; according to the division method of image block nR×2N, the image block is divided into two parts of 1.75M×2M and 0.25M×2M. At this time, the value of log2TrafoSize is reduced by 1, that is, log2TrafoSize=log2CUSize-1.

At this time, the first size is 1<<log2TrafoSize.

Then, the length and width of the transformation block are determined by the determined parameter log2TrafoSize of the transformation block and the division method used by the image block. As shown in Figures 7a and 7b, the 2N×nU and 2N×nD division methods use smaller prediction blocks, in order to ensure that the transformation block does not cross the boundary of the prediction block, the size of the transformation block needs to be made smaller to accommodate The prediction block size, so the transformation block size of these two division methods is (1<<(log2TrafoSize+1))×(1<<(log2TrafoSize-2)), that is, the transformation block size is 2M×0.25M (h=2 , v=0.25).

Similarly, as shown in Figures 7c and 7d, the nL×2N and nR×2N division methods use smaller prediction blocks, and the transform block size needs to become smaller to accommodate the prediction block size, so the two divisions The transform block size of the mode is (1<<(log2TrafoSize-2))×(1<<(log2TrafoSize+1), that is, the transform block size is 0.25M×2M (h=0.25, v=2).

Specifically, when the image block size is 32x32, the value of log2CUSize is 5, and the initial value of log2TrafoSize is 5 as described above. The values of log2TrafoSize are 3, 3, 3, and 3, and the transformation block sizes obtained according to the division method are 32×4, 32×4, 4×32, and 4×32, respectively.

In the embodiment of the present invention, when determining the size of the transformation block, the size of the transformation block is obtained according to the parameters of the transformation block corresponding to the image block and the division method of the image block, so the transformation block suitable for the division method of the image block can be used, and the image compression efficiency is improved. .

Fig. 8 is a flowchart of an image coding method according to an embodiment of the present invention.

801. Obtain at least one candidate transform block size according to parameters of transform block information corresponding to the image block.

Optionally, in an embodiment, the parameters of the transform block may include the number of layers corresponding to the transform block or an indication value of the size of the transform block. At this time, the size of the candidate transform block can be obtained according to the parameter, or the size of the candidate transform block can be obtained according to the parameter and image block information in a manner similar to the above-mentioned embodiments. Similar to the above embodiments, the image block information may include the image block size, the representation value of the image block size, the number of layers of the image block, or the serial number of the image block size.

Different from the foregoing embodiments, this embodiment can obtain one or more candidate transform block sizes for each division method. For example, each of 2N×N, N×2N and N×N divisions may correspond to three candidate transform block sizes M×M, 2M×0.5M and 0.5M×2M.

802. Determine the serial number of at least one candidate transform block size.

The candidate transform block sizes can be directly numbered, for example, starting from a certain transform block size among the candidate transform block sizes (referred to as the first transform block size), they are numbered sequentially as 0, 1, 2....

Optionally, in another embodiment, the sizes of the candidate transformation blocks may be numbered according to the division manner of the image blocks. In other words, for different division methods, the numbers of candidate transform block sizes may be different. At this time, the first transform block size among the candidate transform block sizes may correspond to the division mode, for example, the most likely transform block size or the optimal transform block size in the division mode. For example, for the 2N×N division method, if according to the above-mentioned embodiment, it is considered that the transformation block size 2M×0.5M can reflect the stripe characteristics of the image and is the optimal transformation block size, then the transformation block size 2M×0.5M can be As the first transform block size, the candidate transform block sizes are numbered accordingly.

In an embodiment, the first transformation block size may correspond to a preset number, for example, correspond to the smallest number (such as number "0") or other numbers.

803. Select one of the candidate transform block sizes as the size of the transform block corresponding to the image block, and encode the number of the transform block size.

Optionally, when encoding the number of the size of the transformation block, encode the number of the size of the transformation block into the code stream of the image block, so that the decoding end can parse the number from the code stream to obtain the corresponding transformation block size. In addition, the number value written into the code stream may also be determined according to the selected candidate transform block size and the preset number corresponding to the first transform block size.

Therefore, the embodiment of the present invention selects the transform block size to be used from the candidate transform block sizes when performing image encoding, and encodes the corresponding number of the selected transform block size, so that an appropriate transform block can be selected, improving the Image compression efficiency.

Fig. 9 is a flowchart of an image decoding method according to an embodiment of the present invention.

901. Obtain at least one candidate transform block size according to parameters of a transform block corresponding to an image block.

901 may be similar to 801 in FIG. 8 , so the description will not be repeated.

902. Determine the serial number of at least one candidate transform block size.

902 may be similar to 802 in FIG. 8 , so the description will not be repeated.

903. Obtain the serial number of the transform block size of the transform block.

The operation of the decoding end may correspond to that of the encoding end. For example, if the encoder encodes the number of the size of the transform block into the code stream of the image block, the decoder can parse the code stream to obtain the code of the size of the transform block.

904. Obtain the transform block size according to the serial number of the transform block size.

In 902, the corresponding relationship between the number of the candidate transform block size and the size of the transform block has been obtained, so in 904, the size of the transform block can be obtained according to the number of the transform block size. For example, the transform block size of the transform block may be obtained according to the number of the transform block size obtained by parsing the code stream and the preset number corresponding to the first transform block size.

Therefore, in the embodiment of the present invention, when performing image decoding, the corresponding relationship between the size of the candidate transform block and its number is obtained, so that when the number of the used transform block size is obtained, the size of the transform block can be obtained. The appropriate transformation block of the algorithm improves the image compression efficiency.

The embodiments of the present invention will be described in more detail below in conjunction with specific examples.

In one embodiment, the size of the image block is first determined to be 2M×2M, and 2M is represented by the parameter log2CUSize, and the relationship between the two is 2M=1<<log2CUSize, and the parameter used to identify the basic information of the transformation block size is log2TrafoSize, and the parameter of log2TrafoSize The initial value is log2CUSize. Determine the value of the transformation block log2TrafoSize corresponding to the image block according to the division method of the image block. For example, the method for determining the value of log2TrafoSize of the transformation block corresponding to the image block may be consistent with the method in the foregoing embodiment.

In this embodiment, the 2N×N, N×2N and N×N division methods all have three candidate transform block sizes: M×M, 2M×0.5M and 0.5M×2M, and the encoder traverses each candidate transform block size to determine the optimal transform block size. As shown in Figures 1b, 1c and 1d, the image blocks are divided into 2N×N, N×2N and N×N respectively, and the sizes of the three candidate transform blocks are (1<<log2TrafoSize)×(1<<log2TrafoSize), (1<<(log2TrafoSize+1))×(1<<(log2TrafoSize-1)) and (1<<(log2TrafoSize-1))×(1<<(log2TrafoSize+1)).

The above three candidate transform block sizes are numbered. Wherein, the transform block size corresponding to a reference number is determined according to the division method, for example, in the 2N×N division method, the transform block size corresponding to the reference number 0 is 2M×0.5M. Then, determine the numbers of the other two transform block sizes, such as the 2N×N division method, the numbers corresponding to the M×M and 0.5M×2M transform block sizes are 1 and 2, respectively. Write the number corresponding to the aforementioned optimal transform block size into the code stream.

The purpose of the above encoding process is to determine an optimal transform block size at the encoding end through traversal selection to achieve the best encoding efficiency. At the same time, in order to identify the selected optimal transform block size, it is necessary to number the transform block size and write the number into the code stream. Usually the number is written into the code stream in a variable-length manner, for example, the code '0' uses a 1-bit symbol string '1' to enter the code stream, and the code numbers '1' and '2' use a 2-bit symbol string '00' respectively ' and '01' are written into the code stream. Therefore, allocating the most likely transform block size to the smallest bit symbol string (such as the symbol string corresponding to the number 0 mentioned above) can achieve the effect of saving coding bits and improving coding efficiency. According to the aforementioned analysis, there is a relationship between the division method and the transform block size. For example, the 2N×N division method tends to use the 2M×0.5M transform block size, so the transform block types can be numbered according to the division method.

When the image block uses 2N×N division method, the image block tends to use 2M×0.5M transformation, so the number corresponding to 2M×0.5M is set to 0, and the numbers corresponding to M×M and 0.5M×2M transformation are respectively 1 and 2. When the image block uses the N×2N division method, the image block tends to use 0.5M×2M transformation, so the number corresponding to 0.5M×2M is set to 0, and the numbers corresponding to 2M×0.5M and M×M transformation are respectively 1 and 2. When the image block uses the N×N division method, the image block tends to use M×M transformation, so the number corresponding to M×M is 0, and the numbers corresponding to 2M×0.5M and 0.5M×2M transformation are 1 and 2 respectively .

The decoding end is the reverse process of the encoding end. First, the value of log2TrafoSize of the transformation block corresponding to the image block is determined according to the division method of the image block. The method for determining the value of the log2TrafoSize of the transformation block corresponding to the image block is consistent with the above method. Obtain the candidate transform block size according to the value of log2TrafoSize. Then, analyze the code stream to obtain the number of the transformation block size, and obtain the transformation block size corresponding to the number according to the division method. The transform block size corresponding to the image block is obtained according to the number.

Therefore, the embodiment of the present invention selects the transform block size to be used from the candidate transform block sizes when performing image encoding and decoding, and encodes the corresponding number of the selected transform block size, so that an appropriate transform block can be selected to improve image compression efficiency.

In another embodiment, the size of the image block is first determined to be 2M×2M, 2M is represented by the parameter log2CUSize, the relationship between the two is 2M=1<<log2CUSize, and the parameters used to identify the basic information of the transformation block size are log2TrafoSize, log2TrafoSize The initial value of log2CUSize. Determine the value of the transformation block log2TrafoSize corresponding to the image block according to the division method of the image block.

When the value of log2TrafoSize is determined by the layer-by-layer identification method of the image block, when the transformation block is located at the zeroth layer, the value of log2TrafoSize remains unchanged, that is, log2TrafoSize=log2CUSize; when the transformation block is located at the first layer, the value of log2TrafoSize relative to the zeroth layer The value of layer log2TrafoSize minus 1, that is, log2TrafoSize=log2CUSize-1; when the transformation block is in the second layer, the value of log2TrafoSize relative to the first layer log2TrafoSize minus 1, that is, log2TrafoSize=log2CUSize-2; and so on to obtain log2TrafoSize value.

The basic size parameter TL corresponding to each layer of transformation block is obtained according to the aforementioned value of log2TrafoSize, and the value of TL is equal to (1<<log2TrafoSize). In each layer, the 2N×N, N×2N and N×N division methods have three candidate transform block sizes: TL×TL, 2TL×0.5TL, 0.5TL×2TL, and the encoder traverses each candidate Transform block size to determine optimal transform block size. As shown in Figures 1b, 1c and 1d, the image blocks are divided into 2N×N, N×2N and N×N respectively, and the sizes of the three candidate transform blocks are (1<<log2TrafoSize)×(1<<log2TrafoSize), (1<<(log2TrafoSize+1))×(1<<(log2TrafoSize-1)) and (1<<(log2TrafoSize-1))×(1<<(log2TrafoSize+1)).

The above three candidate transform block sizes are numbered. Wherein, the transformation block size corresponding to a reference number is determined according to the division method, for example, in the 2N×N division method, the transformation block size corresponding to the reference number 0 is 2TL×0.5TL. Then, determine the numbers of the other two transform block sizes, for example, in the 2N×N division mode, the numbers corresponding to the TL×TL and 0.5TL×2TL transform block sizes are 1 and 2, respectively. Write the number corresponding to the aforementioned optimal transform block size into the code stream.

When the image block uses 2N×N division method, the image block tends to use 2TL×0.5TL transformation, so the number corresponding to 2TL×0.5TL is set to 0, and the numbers corresponding to TL×TL and 0.5TL×2TL transformation are respectively 1 and 2. When the image block uses the N×2N division method, the image block tends to use 0.5TL×2TL transformation, so the number corresponding to 0.5TL×2TL is set to 0, and the numbers corresponding to 2TL×0.5TL and TL×TL transformation are respectively 1 and 2. When the image block uses the N×N division method, the image block tends to use TL×TL transformation, so the number corresponding to TL×TL is 0, and the numbers corresponding to 2TL×0.5TL and 0.5TL×2TL transformation are 1 and 2 respectively .

The decoding end is the reverse process of the encoding end. First, the value of log2TrafoSize of the transformation block corresponding to the image block is determined according to the division method of the image block. The method for determining the value of the log2TrafoSize of the transformation block corresponding to the image block is consistent with the above method. Obtain the candidate transform block size according to the value of log2TrafoSize. Then, analyze the code stream to obtain the number of the transformation block size, and obtain the transformation block size corresponding to the number according to the division method. The transform block size corresponding to the image block is obtained according to the number.

Therefore, the embodiment of the present invention selects the transform block size to be used from the candidate transform block sizes when performing image encoding and decoding, and encodes the corresponding number of the selected transform block size, so that an appropriate transform block can be selected to improve image compression efficiency.

In another embodiment, the size of the image block is first determined to be 2M×2M, 2M is represented by the parameter log2CUSize, the relationship between the two is 2M=1<<log2CUSize, and the parameters used to identify the basic information of the transformation block size are log2TrafoSize, log2TrafoSize The initial value of log2CUSize. Determine the value of the transformation block log2TrafoSize corresponding to the image block according to the division method of the image block.

When the value of log2TrafoSize is determined by the layer-by-layer identification method of the image block, when the transformation block is located at the zeroth layer, the value of log2TrafoSize remains unchanged, that is, log2TrafoSize=log2CUSize; when the transformation block is located at the first layer, the value of log2TrafoSize relative to the zeroth layer The value of layer log2TrafoSize minus 1, that is, log2TrafoSize=log2CUSize-1; when the transformation block is in the second layer, the value of log2TrafoSize relative to the first layer log2TrafoSize minus 1, that is, log2TrafoSize=log2CUSize-2; and so on to obtain log2TrafoSize value.

The basic size parameter TL corresponding to each layer of transformation block is obtained according to the aforementioned value of log2TrafoSize, and the value of TL is equal to (1<<log2TrafoSize). In each layer, the 2N×N, N×2N and N×N division methods have three candidate transform block sizes: TL×TL, 2TL×0.5TL, 0.5TL×2TL, and the encoder traverses each candidate Transform block size to determine optimal transform block size. As shown in Figures 1b, 1c and 1d, the image blocks are divided into 2N×N, N×2N and N×N respectively, and the sizes of the three candidate transform blocks are (1<<log2TrafoSize)×(1<<log2TrafoSize), (1<<(log2TrafoSize+1))×(1<<(log2TrafoSize-1)) and (1<<(log2TrafoSize-1))×(1<<(log2TrafoSize+1)).

Number the above candidate transform block sizes, such as (1<<log2TrafoSize)×(1<<log2TrafoSize), (1<<(log2TrafoSize+1))×(1<<(log2TrafoSize-1)) and (1<< (log2TrafoSize-1))×(1<<(log2TrafoSize+1)) corresponds to numbers 0, 1, and 2 respectively, and writes the number corresponding to the aforementioned optimal transform block size into the code stream.

The decoding end is the reverse process of the encoding end. First, the value of log2TrafoSize of the transformation block corresponding to the image block is determined according to the division method of the image block. The method for determining the value of the log2TrafoSize of the transformation block corresponding to the image block is consistent with the above method. Obtain the candidate transform block size according to the value of log2TrafoSize. Then, the code stream is analyzed to obtain the number of the size of the transformation block, and the size of the transformation block corresponding to the image block is obtained according to the number.

Therefore, the embodiment of the present invention selects the transform block size to be used from the candidate transform block sizes when performing image encoding and decoding, and encodes the corresponding number of the selected transform block size, so that an appropriate transform block can be selected to improve image compression efficiency.

Fig. 10 is a block diagram of an apparatus for determining a transform block size according to an embodiment of the present invention. The device 100 in FIG. 10 includes a parameter acquisition unit 101 and a size acquisition unit 102 .

The parameter obtaining unit 101 obtains the parameters of the transformation block corresponding to the image block according to the image block information and the division method of the image block. The size obtaining unit 102 obtains the size of the transform block according to the parameters of the transform block and the division method of the image block.

In the embodiment of the present invention, when determining the size of the transformation block, the size of the transformation block is obtained according to the parameters of the transformation block corresponding to the image block and the division method of the image block, so the transformation block suitable for the division method of the image block can be used, and the image compression efficiency is improved. .

The device 100 in the embodiment of the present invention can execute each process of the foregoing method, and details are not described again to avoid repetition. For example, the parameters of the transformation block obtained by the parameter acquisition unit 101 may include the number of layers corresponding to the transformation block or an indication value of the size of the transformation block.

In addition, the size obtaining unit 102 obtains a transformed block whose horizontal size is larger than the vertical size when the horizontal size of the predicted block is larger than the vertical size; when the vertical size of the predicted block is larger than the horizontal size, the obtained The vertical size of the transform block is larger than the horizontal size, and when the horizontal size of the prediction block is equal to the vertical size, the horizontal size of the obtained transform block is equal to the vertical size.

Fig. 11 is a block diagram of an image encoding device of one embodiment of the present invention. The image encoding device 110 of FIG. 11 includes an acquisition unit 111 , a determination unit 112 , and an encoding unit 113 .

The obtaining unit 111 obtains at least one candidate transform block size according to the parameters of the transform block corresponding to the image block. The determination unit 112 determines the number of at least one candidate transform block size. The encoding unit 113 selects one of the candidate transform block sizes as the size of the transform block corresponding to the image block, and encodes the number of the size of the transform block.

Therefore, the embodiment of the present invention selects the transform block size to be used from the candidate transform block sizes when performing image encoding, and encodes the corresponding number of the selected transform block size, so that an appropriate transform block can be selected, improving the Image compression efficiency.

The device 110 in the embodiment of the present invention can execute each process of the foregoing method, and details are not described again to avoid repetition. For example, the determining unit 112 may number the sizes of the candidate transform blocks according to the division manner of the prediction block.

In addition, the encoding unit 113 may encode the number corresponding to the size of the selected candidate transform block into the code stream of the image block.

Fig. 12 is a block diagram of an image decoding device of an embodiment of the present invention. The image decoding device 120 of FIG. 12 may include an acquisition unit 121 , a determination unit 122 , and a decoding unit 123 .

The obtaining unit 121 obtains at least one candidate transform block size according to the parameters of the transform block corresponding to the image block. The determination unit 122 determines the serial number of at least one candidate transform block size. The decoding unit 123 obtains the number of the transform block size of the transform block, and obtains the transform block size according to the number of the transform block size.

Therefore, in the embodiment of the present invention, when performing image decoding, the corresponding relationship between the size of the candidate transform block and its number is obtained, so that when the number of the used transform block size is obtained, the size of the transform block can be obtained. The appropriate transformation block of the algorithm improves the image compression efficiency.

The device 110 in the embodiment of the present invention can execute each process of the foregoing method, and details are not described again to avoid repetition. For example, the determining unit 122 may number the sizes of the candidate transform blocks according to the division manner of the prediction block.

In addition, the decoding unit 123 can analyze the code stream to obtain the serial number of the transform block size of the transform block.

The operation of the image decoding device 120 of FIG. 12 may correspond to that of the image encoding device of FIG. 11 . For example, if the encoding end encodes the number of the size of the transformation block into the code stream of the image block, the decoding end can parse the code stream to obtain the number of the size of the transformation block.

Those skilled in the art can appreciate that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the present invention.

Those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods may be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or May be integrated into another system, or some features may be ignored, or not implemented. In another point, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, each unit may exist separately physically, or two or more units may be integrated into one unit.

If the functions described above are realized in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the essence of the technical solution of the present invention or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes. .

The above is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Anyone skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present invention. Should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims (40)

1. A method of determining transform block sizes, comprising:

obtaining parameters of a transformation block corresponding to an image block according to the image block information and the division mode of the image block;

obtaining the size of a transform block according to the parameters of the transform block and the division mode of the image block;

the parameter of the transformation block comprises a layer number corresponding to the transformation block or a representation value of the size of the transformation block;

wherein, for an image block with a size of 2N × 2N, when the image block is divided into 2N × N, N × 2N or N × N, the number of layers corresponding to the transform block is 1, and when the image block is divided into 2N × 2N, the number of layers corresponding to the transform block is 0;

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block; or,

the representation value of the transform block size is equal to the representation value of the image block size minus m1, m1 is a positive integer; or,

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block minus m2, and m2 is a positive integer;

wherein the expression values of the image block size and the image block size satisfy the following conditions: image block size 2^log2CUSizeOr image block size equal to 1<<log2CUSize, wherein "<<"represents a leftward shift operation, and log2CUSize is a representative value of the image block size.

2. The method of claim 1, wherein obtaining a transform block size according to the parameters of the transform block and the division of the image block comprises:

when the horizontal direction size of the prediction block is larger than the vertical direction size, the horizontal direction size of the transformation block is larger than the vertical direction size;

when the vertical direction size of the prediction block is larger than the horizontal direction size, the vertical direction size of the transformation block is larger than the horizontal direction size;

when the horizontal direction size of the prediction block is equal to the vertical direction size, the horizontal direction size of the transform block is equal to the vertical direction size.

3. The method of claim 2,

when the prediction block type is 2NxN, the horizontal direction size of the transformation block is 4 times of the vertical direction size; or,

when the prediction block type is N × 2N, the vertical direction size of the transform block is 4 times the horizontal direction size; or,

when the prediction block type is nxn or 2 nx2N, the vertical dimension of the transform block is equal to the horizontal dimension; or,

when the prediction block type is 2NxnU or 2NxnD, the horizontal direction size of the transformation block is 4 times of the vertical direction size; or,

when the prediction block type is nL multiplied by 2N or nR multiplied by 2N, the vertical direction size of the transformation block is 4 times of the horizontal direction size;

wherein, the 2NxnU and the 2NxnD represent that the current image block is divided into two rectangular sub image blocks with different sizes;

nLx2N and nRx2N indicate that the current image block is divided into two unequal-sized rectangular sub-image blocks left and right.

4. The method of claim 2,

when the prediction block type is 2NxnU or 2NxnD, the horizontal direction size of the transformation block is 8 times of the vertical direction size; or,

when the prediction block type is nL multiplied by 2N or nR multiplied by 2N, the vertical direction size of the transformation block is 8 times of the horizontal direction size;

wherein, the 2NxnU and the 2NxnD represent that the current image block is divided into two rectangular sub image blocks with different sizes;

nLx2N and nRx2N indicate that the current image block is divided into two unequal-sized rectangular sub-image blocks left and right.

5. The method of claim 1, wherein obtaining a transform block size according to the parameters of the transform block and the division of the image block comprises:

obtaining a first size of a transform block according to the parameter;

obtaining the transform block size according to the first size.

6. The method of claim 5, wherein said obtaining the transform block size according to the first size comprises:

the transform block has a horizontal dimension equal to h times the first dimension, and a vertical dimension equal to v times the first dimension, where h and v are greater than 0.

7. The method of claim 6,

h is greater than v when the horizontal dimension of the prediction block is greater than the vertical dimension;

h is less than v when the horizontal dimension of the prediction block is less than the vertical dimension;

when the horizontal direction size of the prediction block is equal to the vertical direction size, h is equal to v.

8. The method of claim 7,

when the prediction block type is 2N × N, h is 2, and v is 0.5; or,

when the prediction block type is N × 2N, h is 0.5 and v is 2; or,

when the prediction block type is nxn, h is 1, and v is 1; or,

when the prediction block type is 2N × 2N, h is 2 and v is 2; or,

when the prediction block type is 2nxnu or 2nxnd, h is 2 and v is 0.5; or,

when the prediction block type is nL × 2N or nR × 2N, h is 0.5, v is 2,

wherein N is a positive integer, N is greater than 0 and less than N;

wherein, the 2NxnU and the 2NxnD represent that the current image block is divided into two rectangular sub image blocks with different sizes;

nLx2N and nRx2N indicate that the current image block is divided into two unequal-sized rectangular sub-image blocks left and right.

9. The method of claim 7,

when the prediction block type is 2nxnu or 2nxnd, h is 2 and v is 0.25; or,

when the prediction block type is nL × 2N or nR × 2N, h is 0.25, v is 2,

wherein N is a positive integer, N is greater than 0 and less than N;

wherein, the 2NxnU and the 2NxnD represent that the current image block is divided into two rectangular sub image blocks with different sizes;

nLx2N and nRx2N indicate that the current image block is divided into two unequal-sized rectangular sub-image blocks left and right.

10. The method according to claim 1, wherein before obtaining the parameters of the transform block corresponding to the image block according to the image block information and the division manner of the image block, the method further comprises:

obtaining a dividing mark value according to the type of the prediction block and a layering mark value, wherein the layering mark value is used for indicating whether a layer-by-layer marking mode is adopted by the transformation block;

and determining whether to determine the size of the transform block according to the division mode of the image block based on the division flag value.

11. The method of any one of claims 1 to 10, wherein the image block information includes an image block size, a value indicating an image block size, or a number of an image block size.

12. An image encoding method, comprising:

obtaining at least one candidate transformation block size according to the parameters of the transformation block corresponding to the image block; the parameter of the transformation block comprises a layer number corresponding to the transformation block or a representation value of the size of the transformation block;

determining a number of the at least one candidate transform block size;

selecting one of the candidate transform block sizes as a transform block size of a transform block corresponding to the image block, and encoding a number of the size of the transform block;

the determining the number of the at least one candidate transform block size comprises:

according to the way in which the image blocks are divided,

determining a first transform block size of the at least one candidate transform block size, numbering the candidate transform block sizes according to the first transform block size;

wherein, for an image block with a size of 2N × 2N, when the image block is divided into 2N × N, N × 2N or N × N, the number of layers corresponding to the transform block is 1, and when the image block is divided into 2N × 2N, the number of layers corresponding to the transform block is 0;

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block; or,

the representation value of the transform block size is equal to the representation value of the image block size minus m1, m1 is a positive integer; or,

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block minus m2, and m2 is a positive integer;

wherein the representation values of the transform block size and the transform block size satisfy the following conditions: transform block size 2^{log2TrafoSize}Or transform block size 1<<log2TrafoSize, wherein "<<"represents a left shift operation, and log2TrafoSize is a representative value of the transform block size.

13. The method of claim 12, wherein said numbering the candidate transform block sizes according to the first transform block size comprises: the first transform block size is corresponding to a preset number.

14. The method of claim 13, wherein the preset number is number 0.

15. The method of claim 12, wherein said encoding the number of sizes of the transform blocks comprises: and coding the serial number of the size of the transformation block into the code stream of the image block.

16. The method of claim 13, wherein said encoding a number of sizes of said transform blocks comprises: and determining the number value of the written code stream according to the selected candidate transformation block size and the preset number corresponding to the first transformation block size.

17. The method of claim 12, wherein obtaining at least one candidate transform block size based on parameters of a transform block corresponding to the image block comprises:

and obtaining the at least one candidate transform block size according to the parameters and the image block information.

18. The method of claim 17, wherein the image block information comprises a block size of an image, a representative value of the block size of the image, the number of layers of the image block, or the number of the size of the image block.

19. An image decoding method, comprising:

obtaining at least one candidate transformation block size according to the parameters of the transformation block corresponding to the image block; the parameter of the transformation block comprises a layer number corresponding to the transformation block or a representation value of the size of the transformation block;

determining a number of the at least one candidate transform block size;

obtaining the number of the size of a transformation block of the transformation block;

obtaining the size of the transformation block according to the serial number of the size of the transformation block;

the determining the number of the at least one candidate transform block size comprises:

according to the way in which the image blocks are divided,

determining a first transform block size of the at least one candidate transform block size, numbering the candidate transform block sizes according to the first transform block size;

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block; or,

the representation value of the transform block size is equal to the representation value of the image block size minus m1, m1 is a positive integer; or,

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block minus m2, and m2 is a positive integer;

wherein, for an image block with a size of 2N × 2N, when the image block is divided into 2N × N, N × 2N or N × N, the number of layers corresponding to the transform block is 1, and when the image block is divided into 2N × 2N, the number of layers corresponding to the transform block is 0;

wherein the representation values of the transform block size and the transform block size satisfy the following conditions: transform block size 2^{log2TrafoSize}Or transform block size 1<<log2TrafoSize, wherein "<<"represents a left shift operation, and log2TrafoSize is a representative value of the transform block size.

20. The method of claim 19, wherein said numbering the candidate transform block sizes according to the first transform block size comprises: the first transform block size is corresponding to a preset number.

21. The method of claim 20, wherein the preset number is number 0.

22. The method of claim 19, wherein said obtaining the number of transform block sizes for the transform block comprises:

and analyzing the code stream to obtain the serial number of the transform block size of the transform block.

23. The method of claim 20, wherein obtaining a transform block size based on the number of transform block sizes comprises:

and obtaining the transformation block size of the transformation block according to the number of the transformation block size of the transformation block and the preset number corresponding to the first transformation block size.

24. The method of claim 19, wherein obtaining at least one candidate transform block size based on parameters of a transform block corresponding to the image block comprises:

and obtaining the at least one candidate transform block size according to the parameters and the image block information.

25. The method of claim 24, wherein the image block information comprises a block size of an image, a representative value of the block size of the image, the number of layers of the image block, or the number of the size of the image block.

26. An apparatus for determining transform block sizes, comprising:

the parameter acquisition unit is used for acquiring parameters of a transformation block corresponding to the image block according to the image block information and the division mode of the image block; the parameter of the transformation block comprises a layer number corresponding to the transformation block or a representation value of the size of the transformation block;

the size obtaining unit is used for obtaining the size of the transformation block according to the parameters of the transformation block and the division mode of the image block;

wherein, for an image block with a size of 2N × 2N, when the image block is divided into 2N × N, N × 2N or N × N, the number of layers corresponding to the transform block is 1, and when the image block is divided into 2N × 2N, the number of layers corresponding to the transform block is 0;

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block; or,

the representation value of the transform block size is equal to the representation value of the image block size minus m1, m1 is a positive integer; or,

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block minus m2, and m2 is a positive integer;

wherein the representation values of the transform block size and the transform block size satisfy the following conditions: transform block size 2^{log2TrafoSize}Or transform block size 1<<log2TrafoSize, wherein "<<"represents a left shift operation, and log2TrafoSize is a representative value of the transform block size.

27. The apparatus of claim 26, wherein the size obtaining unit obtains the horizontal direction size of the transform block to be larger than the vertical direction size when the horizontal direction size of the prediction block is larger than the vertical direction size, obtains the vertical direction size of the transform block to be larger than the horizontal direction size when the vertical direction size of the prediction block is larger than the horizontal direction size, and obtains the horizontal direction size of the transform block to be equal to the vertical direction size when the horizontal direction size of the prediction block is equal to the vertical direction size.

28. An image encoding device characterized by comprising:

the device comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring at least one candidate transformation block size according to the parameter of the transformation block corresponding to the image block; the parameter of the transformation block comprises a layer number corresponding to the transformation block or a representation value of the size of the transformation block;

an encoding unit configured to select one of the candidate transform block sizes as a size of a transform block corresponding to the image block, and encode a number of the size of the transform block;

a determining unit, configured to determine a first transform block size of the at least one candidate transform block size according to a partition manner of a prediction block, and number the candidate transform block size according to the first transform block size;

wherein, for an image block with a size of 2N × 2N, when the image block is divided into 2N × N, N × 2N or N × N, the number of layers corresponding to the transform block is 1, and when the image block is divided into 2N × 2N, the number of layers corresponding to the transform block is 0;

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block; or,

the representation value of the transform block size is equal to the representation value of the image block size minus m1, m1 is a positive integer; or,

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block minus m2, and m2 is a positive integer;

wherein the representation values of the transform block size and the transform block size satisfy the following conditions: transform block size 2^{log2TrafoSize}Or transform block size 1<<log2TrafoSize, wherein "<<"represents a left shift operation, and log2TrafoSize is a representative value of the transform block size.

29. The image encoding apparatus of claim 28, wherein the encoding unit is configured to encode a number corresponding to the selected candidate transform block size into a code stream of the image block.

30. An image decoding apparatus characterized by comprising:

the device comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring at least one candidate transformation block size according to the parameter of the transformation block corresponding to the image block; the parameter of the transformation block comprises a layer number corresponding to the transformation block or a representation value of the size of the transformation block;

a decoding unit, configured to obtain the number of the transform block size of the transform block, and obtain the transform block size according to the number of the transform block size;

the determining unit is used for determining a first transform block size in the at least one candidate transform block size according to the dividing mode of the prediction block, and numbering the candidate transform block size according to the first transform block size;

wherein, for an image block with a size of 2N × 2N, when the image block is divided into 2N × N, N × 2N or N × N, the number of layers corresponding to the transform block is 1, and when the image block is divided into 2N × 2N, the number of layers corresponding to the transform block is 0;

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block; or,

the representation value of the transform block size is equal to the representation value of the image block size minus m1, m1 is a positive integer; or,

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block minus m2, and m2 is a positive integer;

wherein the representation values of the transform block size and the transform block size satisfy the following conditions: transform block size 2^{log2TrafoSize}Or transform block size 1<<log2TrafoSize, wherein "<<"represents a left shift operation, and log2TrafoSize is a representative value of the transform block size.

31. The image decoding device according to claim 30, wherein the decoding unit is configured to parse a code stream to obtain a number of a transform block size of the transform block.

32. A method of determining transform block sizes, comprising:

obtaining a division parameter according to the division mode of the image block and the parameter of the transformation block, and obtaining the size of the transformation block according to the division parameter; the parameter of the transformation block comprises a layer number corresponding to the transformation block or a representation value of the size of the transformation block;

the obtaining of the partition parameter according to the partition mode of the image block and the parameter of the transform block comprises:

when the image block is divided into 2NxN or Nx2N, and when the number of division layers corresponding to the transform block is 0, the division parameter value is 1, where 2NxN denotes that the current image block is divided into two sub image blocks with equal sizes, and Nx2N denotes that the current image block is divided into two sub image blocks with equal sizes; or,

when the image block is divided into 2NxnU, 2NxnD, nLx2N, or nRx2N, and the number of division layers corresponding to the transform block is 0 or 1, the value of the division parameter is 1,

wherein, the 2NxnU and the 2NxnD represent that the current image block is divided into two rectangular sub image blocks with different sizes; nLx2N and nRx2N represent that the current image block is divided into two unequal-sized rectangular sub-image blocks on the left and right;

wherein, for an image block with a size of 2N × 2N, when the image block is divided into 2N × N, N × 2N or N × N, the number of layers corresponding to the transform block is 1, and when the image block is divided into 2N × 2N, the number of layers corresponding to the transform block is 0;

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block; or,

the representation value of the transform block size is equal to the representation value of the image block size minus m1, m1 is a positive integer; or,

the representation value of the size of the transformation block is equal to the representation value of the size of the image block minus the number of layers corresponding to the transformation block minus m2, and m2 is a positive integer;

wherein the representation values of the transform block size and the transform block size satisfy the following conditions: transform block size 2^{log2TrafoSize}Or transform block size 1<<log2TrafoSize, wherein "<<"represents a left shift operation, and log2TrafoSize is a representative value of the transform block size.

33. The method of claim 32, wherein the parameter of the transform block includes a value indicative of a number of division layers and a size of the transform block corresponding to the transform block.

34. The method of claim 32, wherein a maximum transform block layer number corresponding to the transform block is obtained according to the partition parameter.

35. The method of claim 32, wherein the obtaining of the division parameter according to the division of the image block and the parameter of the transform block, and the obtaining of the transform block size according to the division parameter further comprises:

when the image block is divided into 2NxnU, 2NxnD, nLx2N or nRx2N and the dividing parameter value is 1, obtaining a parameter for identifying whether the image block is divided further;

and obtaining the maximum conversion block layer number corresponding to the conversion block according to the dividing parameter and the parameter of whether to further divide.

36. The method of claim 32, wherein whether the transform block is divided is determined according to the division parameter, and a transform block division flag is obtained according to the division parameter;

and obtaining the size of the transformation block according to the transformation block division mark.

37. The method of claim 35, wherein the number of division layers corresponding to the transform block is increased by 1 when the transform block division flag is 1.

38. The method of claim 36 or 37, wherein the encoding side does not encode the transform block division flag when the number of division layers is greater than or equal to a maximum transform block layer number corresponding to the transform block.

39. The method of claim 36 or 37, wherein the decoding end does not decode the transform block division flag when the number of division layers is greater than or equal to a maximum transform block layer number corresponding to the transform block.

40. The method of claim 35, wherein the representation value of the transform block size is decreased by 1 when the transform block division flag is 1.